The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute local Python utility scripts provided within the skill package to manage content indices and retrieve materials. Specifically, it calls python3 scripts/rebuild_published_index.py to synchronize file lists and surface_candidates.py --kw to search the user's content vault during the assembly phase. While these are vendor-provided resources, instruction-driven shell execution represents a known vulnerability surface.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted external data, including user-supplied topics and content units retrieved from a local vault, creating a potential surface for indirect injection attacks.
Ingestion points: User-provided topics (Phase 0) and materials stored in the local vault (Phase 2).
Boundary markers: The skill does not define specific delimiters or instructions to ignore embedded commands within the interpolated content.
Capability inventory: The skill has the capability to execute local shell commands (Python scripts).
Sanitization: No explicit content filtering or sanitization mechanisms are described for the ingested data before it is processed by the agent.

aikey-script